JUNE 2006
1647
BASTIN ET AL.
On the Interaction between Sea Breeze and Summer Mistral at the Exit of the
Rhône Valley
SOPHIE BASTIN,* PHILIPPE DROBINSKI,* VINCENT GUÉNARD,⫹ JEAN-LUC CACCIA,⫹
BERNARD CAMPISTRON,# ALAIN M. DABAS,@ PATRICIA DELVILLE,& OLIVER REITEBUCH,**
CHRISTIAN WERNER**
AND
* Institut Pierre Simon Laplace/Service d’Aéronomie, Paris, France
⫹ Laboratoire de Sondages Electromagnétiques de l’Environnement Terrestre, La Garde, France
# Laboratoire d’Aérologie, Toulouse, France
@ Centre National de Recherches Météorologiques, Météo-France, Toulouse, France
& Division Technique/Institut National des Sciences de l’Univers, Meudon, France
** Deutsches Zentrum für Luft- und Raumfahrt, Wessling, Germany
(Manuscript received 19 November 2004, in final form 28 July 2005)
ABSTRACT
The three-dimensional structure and dynamics of the combination of the sea breeze and the mistral at the
Rhône Valley exit, in southeastern France, have been investigated experimentally and numerically on 22
June 2001. The mistral refers to a severe northerly wind that develops along the Rhône Valley. The exit of
this valley is located near the Mediterranean Sea where sea-breeze circulation often develops. The sea
breeze and the mistral coexist this day because of the weakness of this mistral event.
The event was documented in the framework of the Expérience sur Site pour Contraindre les Modèles
de Pollution Atmosphérique et de Transport d’Emissions (ESCOMPTE) field experiment. Several important data sources are used (airborne Doppler lidar, UHF wind profilers, radiosoundings, and surface
stations) as well as nonhydrostatic mesoscale simulations.
This paper examines the various mechanisms that drive the time and spatial variability of the mistral and
the sea breeze in various regions of the Rhône Valley. In the morning, the sea breeze penetrates inland near
the western side of the Rhône Valley then moves back because of the reinforcement of the mistral flow
caused by the deepening of the leeward surface low due to convection at noon. At midday, the sea breeze
penetrates inland in the middle of the Rhône Valley only. In contrast to pure sea-breeze episodes when the
sea breeze can extend inland over a horizontal range of more than 150 km, the presence of the mistral
prevents the sea breeze from penetrating more than 40 km onshore. In the late afternoon, the sea breeze
reaches the eastern side of the Rhône Valley but over a smaller horizontal range because of higher local
topography and because the mistral is more intense in this part of the Rhône Valley.
The situations of sea-breeze–mistral interactions can have a severe impact on regional air quality. Indeed,
the southerly sea breeze, which advects toward the countryside the pollutants emitted from the large coastal
city of Marseille, France, and its industrialized suburbs, cannot penetrate far inland because of the mistral
blowing in the opposite direction. This leads to the stagnation of the pollutants near the area of emission
that is also the most densely inhabited area of the region (over one million inhabitants).
1. Introduction
The Expérience sur Site pour Contraindre les
Modèles de Pollution Atmosphérique et de Transport
d’Emissions (ESCOMPTE) program (Cros et al. 2004),
conducted in June and July 2001, aims at improving the
Corresponding author address: Dr. Sophie Bastin, Earth Observing Laboratory, NCAR, P.O. Box 3000, Boulder, CO 803073000.
E-mail: [email protected]
© 2006 American Meteorological Society
MWR3116
understanding and the forecast of pollutant behavior
during photochemical episodes in Provence, southeastern France (region shown in Fig. 1). Indeed, this region
presents a high occurrence of photochemical pollution
events. On the shore of Provence, the large city of
Marseille and its industrialized suburbs (oil plants in
the Fos-Berre area) are sources of pollutant emission
that cause frequent and hazardous pollution episodes,
especially in summer when intense solar heating enhances the photochemical activity. From the dynamical
characteristics of the area, the atmospheric circulation
1648
MONTHLY WEATHER REVIEW
VOLUME 134
FIG. 1. (a) Map of France with the topography shaded in gray when higher than 500 m above sea level.
The rectangle displays the large domain (hereafter called domain 1) of the Méso-NH simulations. (b)
Domain 1 of the simulation with its nested smaller domain (hereafter called domain 2) in the rectangle.
The acronyms NIM, LYO, and VAL stand for the city names Nîmes, Lyon, and Valence, respectively.
(c) Domain 2 of the Méso-NH simulation. The solid line corresponds to a segment of the flight track of
the DLR Falcon 20 carrying the Doppler lidar WIND on 22 Jun 2001, along which a cross section of the
wind field is shown in Fig. 7 (A and B indicate the limits of the segment). The flight altitude is 6.5 km
and the aircraft speed is about 170 m s⫺1. The dots indicate the locations of the operational meteorological surface stations operated by Météo-France. The acronyms AIX, MAR, MRS, NIM, and STC
correspond to the city names Aix-en-Provence, Marignane, Marseille, Nîmes, and Saint Chamas, respectively.
over this site is highly influenced by orography. On the
large scale, the French Alps (highest elevation, 4807 m)
and Massif Central (highest elevation, 1885 m) at times
reinforce a strong northerly flow, called mistral, in the
Rhône Valley (e.g., Guénard et al. 2005a; Drobinski et
al. 2005). On a smaller scale, the ranges of smaller
mountains parallel to the coastline (i.e., Sainte Baume,
Sainte Victoire, Luberon, Ventoux) tend to channel the
air masses in a west–east manner (Bastin et al. 2005b).
Finally, the sea–land contrasts induce the sea breeze
during daytime, with advection of marine air masses as
far as 100 km inland (e.g., Bastin et al. 2005b; Bastin
and Drobinski 2005a).
During pollution episodes, the sea breeze can export
inland the pollutants emitted from the Marseille urban
area and the Fos-Berre plants, and the countryside is
often more polluted than the area immediately sur-
rounding the pollutant sources. Table 1 illustrates this
point. Indeed, it shows that the ozone concentration is
low at Marseille under pure sea-breeze conditions (25
and 26 June, 3 and 4 July), while pollution episodes
have been predicted and occurred these days. The reason is that the ozone plume is advected farther inland
(Menut et al. 2005).
In Provence, the sea breeze shares its occurrence
with the mistral. The mistral refers to a severe northerly
wind that develops along the Rhône Valley. The mistral
genesis is preconditioned by cyclogenesis over the Gulf
of Genoa and the passage of a trough through France.
When the synoptic northerly/northwesterly flow impinges on the Alpine range, it splits and is deflected
west by the Coriolis force as well as the pressure
buildup on the northern edge of the range. As the flow
experiences channeling in the Rhône Valley separating
JUNE 2006
TABLE 1. Ozone concentration measured at Marseille and averaged between 1000 and 1600 UTC, as a function of the type of
atmospheric circulation.
Date
17
18
19
21
22
23
24
25
26
28
1
3
4
Jun 2001
Jun 2001
Jun 2001
Jun 2001
Jun 2001
Jun 2001
Jun 2001
Jun 2001
Jun 2001
Jun 2001
Jul 2001
Jul 2001
Jul 2001
1649
BASTIN ET AL.
Type of atmospheric flow
Ozone
concentration
(␮g m⫺3)
Mistral
Mistral
Mistral
Mistral ⫹ sea breeze
Mistral ⫹ sea breeze
Mistral ⫹ sea breeze
Mistral (end) ⫹ sea breeze
Sea breeze
Sea breeze
Mistral
Mistral
Sea breeze
Sea breeze
76
81
84
140
115
120
100
62
92
88
101
100
91
the French Alps, to the east, from the Massif Central, to
the west, by a gap of 200 km long and 60 km wide, it is
substantially accelerated, giving rise to the mistral. The
mistral is often strong enough to inhibit the sea-breeze
flow (Arritt 1993) and to prevent any sea-breeze development along the coastline. It is even frequently observed to extend as far as a few hundreds of kilometers
from the coast (Jansá 1987). It is thus associated with
low pollution levels in Provence as it advects the pollutants away from their sources of emission over
the Mediterranean. Table 1 shows that during the
ESCOMPTE experiment, the mistral events correspond to low ozone concentration over Marseille, with
one exception of the 1 July 2001 mistral case when the
ozone concentration was higher than usual because
high-level tropospheric ozone was incorporated within
the mistral flow (Corsmeier et al. 2005).
During the ESCOMPTE field experiment, between
21 and 23 June 2001, a mistral event occurred, but its
weak intensity allowed the sea breeze to break through
during daytime. We believe that because of the adverse
mistral flow, the sea breeze could not penetrate far
inland, making the pollutants stagnate close to the
coastline. This hypothesis is investigated in this study
but high ozone concentration levels measured at
Marseille (Table 1) tend to confirm this assumption.
The paper focuses on 22 June 2001 since it is the most
stationary day of the 3-day period. Specifically, this paper analyzes the dynamic processes causing the unsteady and inhomogeneous development of the mistral
and the sea breeze at the small scale in the Rhône
Valley.
The objectives of the paper are addressed using both
the dataset collected during the ESCOMPTE experi-
ment and the nonhydrostatic mesoscale model MésoNH (Lafore et al., 1998). We make use of the data
from the operational meteorological surface station
network, the radiosoundings, the UHF wind profilers,
and an airborne Doppler lidar. In the framework of
ESCOMPTE, the UHF wind profilers were used to investigate the dynamical processes that drive the time
variability of the mistral (Caccia et al. 2004; Guénard et
al. 2005a), or that explain the complex layered structure
of ozone concentration near the coast during seabreeze events (Delbarre et al. 2005; Puygrenier et al.
2005). The airborne Doppler lidar proved to be a
proper instrument to investigate the nature of the mistral (Drobinski et al. 2005) or the structure of the sea
breeze (Bastin et al. 2005b) because of its ability to map
the wind field in three dimensions.
After the introduction in section 1, the instrument
setup and the numerical model used in this study are
described in section 2. In section 3, the mistral–seabreeze event is described. Section 4 is dedicated to the
evaluation of the model performance in reproducing
this event. Section 5 examines the mechanisms driving
the unsteady and inhomogeneous aspects of the flow
structure at the Rhône Valley exit. Section 6 concludes
the study.
2. Measurements and model
a. Observations
During the ESCOMPTE experiment, a wide range of
instruments was deployed around Marseille, leading to
a dense network of observations available from Doppler and ozone lidars, wind profilers, sodars, radiosoundings, and meteorological surface stations (see the details in Cros et al. 2004). The aim of this experiment was
to study the role of land–sea-breeze circulations on air
pollution transport at local to regional scale. The 22
June 2001 mistral case is the second day of intensive
observing period (IOP) 2a. In the present study, we
mainly use the measurements from the airborne Doppler lidar, the wind profilers, the operational radiosoundings, and the synoptic meteorological stations of
Météo-France. The complementarity of the data provided by these various instruments is an essential aspect
of this study. The locations of the operational meteorological surface stations are shown with dots in Fig. 1c.
1) RADIOSONDES
AND IN SITU SURFACE STATIONS
On 22 June 2001, operational radiosondes were released every 12 h from Lyon and Nîmes (see Fig. 1b).
At the surface, the operational meteorological surface
station network gave access to the surface thermody-
1650
MONTHLY WEATHER REVIEW
namical field (wind, temperature, humidity, pressure).
In addition to this network, few stations providing radiative flux measurements were deployed for the campaign.
2) THE UHF
WIND PROFILERS
Four UHF wind profilers, manufactured by Degréane, were deployed during ESCOMPTE: they were
located at Saint Chamas (STC), Aix-les-Milles near
Aix-en-Provence (AIX), Marseille (MRS), and Marignane (MAR) (see Fig. 1c). The measurements consist
of the time evolution of the vertical profiles of the three
wind components thanks to one vertical beam and two
or four oblique beams (depending upon the radar),
slanted at an off-zenith angle of 17°, the half-power
beamwidth being 8.5°. They work with a frequency of
1238 MHz (⯝30 cm wavelength), and with a peak
power of 4 kW. Returned echoes are due to the air
refractive index fluctuations advected by the wind. The
wind velocity is estimated from the frequency corresponding to the mean Doppler shift obtained in the
radar echo. The data quality control and processing are
carried out through a consensus algorithm based on
time and height continuity of measured spectra. The
consensus works over a 60-min period providing a wind
profile each 15-min from a height of 100–300 m up to
2500–4000 m AGL. The vertical resolution is typically
75–150 m. The errors on the horizontal (vertical) wind
measurements are typically 1–2 m s⫺1 (0.25–0.5 m s⫺1).
3) THE
AIRBORNE
DOPPLER
LIDAR
WIND
The French–German airborne Doppler lidar, the
Wind Infrared Doppler Lidar (WIND; Werner et al.
2001), was operated on the 22 June 2001 mistral case
between 1434 and 1653 UTC. The lidar is operated at
10.6 ␮m in the infrared spectral region and was on
board the Falcon 20 of the Deutsches Zentrum für
Luft- und Raumfarht (DLR; Werner et al. 2001). Between 1635 and 1653 UTC, it flew along track B–A,
shown in Fig. 1c, at an altitude of 6.5 km with an aircraft
ground speed of around 170 m s⫺1. At 10.6-␮m wavelength, the lidar signals are sensitive to micronic aerosols, which are excellent tracers of the dynamics in troposphere, therefore making WIND a relevant tool for
the study of planetary boundary layer (PBL) dynamics
in complex terrain (e.g., Reitebuch et al. 2003; Bastin et
al. 2005b; Drobinski et al. 2005). The Doppler lidar
WIND provides wind radial velocity measurements
along the line of sight (LOS) of the transmitted laser
beam. The wind profile is obtained by conically scanning the LOS around the vertical axis with a fixed angle
of 30° from nadir. The profile of the three-dimensional
wind vector was calculated from the profiles of the LOS
VOLUME 134
wind speeds using a velocity–azimuthal display (VAD)
technique (Caya and Zawadzki 1992). A full scan revolution of the line of sight takes 20 s, leading to a horizontal resolution of about 3.4 km between vertical profiles of the wind vector. The vertical resolution of the
wind profiles is 250 m, and the accuracy of the horizontal wind velocity is around 1 m s⫺1 (Reitebuch et al.
2001).
b. Model
The numerical simulation was conducted with the
Méso-NH model that solves the nonhydrostatic and
anelastic equation system (Lafore et al. 1998). Two interactively nested model domains are used, the horizontal mesh size being 9 and 3 km, respectively. Domains 1 (coarse domain) and 2 (fine domain) are centered at 43.7°N, 4.6°E and cover an area of 450 km ⫻
450 km and 228 km ⫻ 246 km, respectively (see Fig. 1).
Domain 1 covers the Rhône Valley, the western Alps,
and the Massif Central. Domain 2 covers the Rhône
Valley delta. The vertical grid is made of 50 levels with
a mesh stretched between 60 and 600 m. To insure a
good description of the PBL, 12 levels are taken below
1000 m. The size of the first nine meshes is less than
100 m. Above 8000 m, the mesh size is constant and set
to 600 m. The top of the domain is located at 18 000 m.
A complete set of physics parameterization is used. The
turbulence scheme is unidimensional on the vertical
and is based upon the physical mixing length of
Bougeault and Lacarrère (1989). The convection
scheme is described in Bechtold et al. (2001) and is used
in model domain 1. In model domain 2, this parameterization is not needed since convection is resolved explicitly at such high resolution. The radiation schemes,
as well as the parameterization of land surface processes are described in Morcrette (1991) and Noilhan
and Planton (1989), respectively. The initial and coupling fields were generated by first interpolating the
European Centre for Medium-Range Weather Forecasts (ECMWF) analyses available every 6 h on a
0.5° ⫻ 0.5° latitude–longitude grid to the model grid.
The initialization date is 21 June 2001 at 0600 UTC and
the simulation ends on 23 June 2001 at 0000 UTC.
3. The ESCOMPTE IOP2a mistral/sea-breeze
event
a. Synoptic environment
The 22 June 2001 mistral event is featured by a northwesterly flow over France resulting from an anticyclone
over western France and a low pressure system over
northern Europe (Fig. 2). On 21 June 2001 at 1200 UTC
(Fig. 2a), a high-surface-pressure zone (1022 hPa) is
JUNE 2006
1651
BASTIN ET AL.
FIG. 2. Synoptic situation in 12-h intervals on (a) 1200 UTC 21 Jun 2001, (b) 0000 UTC 22 Jun 2001, (c) 1200 UTC
22 Jun 2001, and (d) 0000 UTC 23 Jun 2001 from ECMWF analyses. The mean sea level pressure and 500-hPa
geopotential heights are shown with solid (the contour interval is 2 hPa) and thick dashed (the contour is 50 m)
lines, respectively.
centered over western Ireland and extends over northwestern France. In southeastern France, no significant
pressure gradient prevails. This situation leads to a
northwesterly surface flow and no well established flow
over the western Mediterranean. Nevertheless, a weak
surface low is generated (1010–1012 hPa) in the wake of
the Alps ridge, similar to a föhn knee (see example in
Drobinski et al. 2003), which marks the existence of a
weak mistral wind channeled in the Rhône Valley. On
22 June 2001 at 0000 UTC (Fig. 2b), the high-surfacepressure zone that was located over western Ireland
moves northward, contributing to the weakening of the
surface winds upstream the Massif Central and the
Alps. In southeastern France, the Genoa cyclone
strengthens (however, remaining weak with respect to
other mistral events; see, e.g., Guénard et al. 2005a;
Drobinski et al. 2005) extending from the Pyrénées to
the Pô Valley. On 22 June 2001 at 1200 UTC (Fig. 2c),
the Genoa cyclone is removed by a föhn knee associated with the surface pressure gradient, which is favorable to large cyclonic curvature of the mistral at the exit
of the Rhône Valley. The situation on 23 June 2001 at
0000 UTC is similar to that on 22 June 2001 at 0000
UTC (Fig. 2d). This mistral event differs significantly
from that of 28 June 2001 described in Drobinski et al.
(2005) because the weakness of the Genoa cyclone. The
comparatively small induced pressure gradient between
the windward side and the leeward side of the Alps
generates weaker winds, which may explain why the sea
breeze can break through on that day.
b. Flow structure in the Rhône Valley
1) NEAR-SURFACE
FLOW PATTERN
At 0900 UTC, the flow pattern at the exit of the
Rhône Valley is shown in Fig. 3, which displays (left)
measured and (right) simulated surface temperature
and wind fields [the temperature is represented instead
of the potential temperature since the surface pressure
is only measured at very few stations, but only the measurements made below 500 m above sea level (ASL)
are shown]. Figure 3a shows that the mistral blows over
1652
MONTHLY WEATHER REVIEW
FIG. 3. The 10-m wind and 2-m temperature fields from (left) meteorological surface stations and (right) Méso-NH simulations at (a), (b) 0900, (c), (d) 1100, (e), (f) 1400, and (g), (h)
1700 UTC 22 Jun 2001. The simulated data used for this figure correspond to the data
interpolated at the locations of the surface stations. The topography mask corresponds to
topographical elements higher than 500 m ASL. The arrows indicate the wind direction and
their scale indicates the intensity. The isolines indicate the temperature. Contour interval is
3°C from 23° to 38°C. The sea-breeze front location is indicated by a thick dashed line.
VOLUME 134
JUNE 2006
1653
BASTIN ET AL.
the nearly entire area, except for two regions: on the
western side of the domain (longitude lower than 4°E)
and on the eastern side of the domain (longitude
greater than 5.6°E). Indeed, these are two sheltered
area in the wakes trailing downstream the Massif Central and the western Alps, respectively. The mistral
brings cold and dry continental air (Bastin et al. 2005a)
and inhibits temperature rise. The temperature gradient between sea and land is thus damped and contributes to delay the sea-breeze onset in the region where
the mistral blows. On the contrary, in the sheltered
area, the temperature naturally increases with radiative
heating (in the sheltered area, the temperature increases from 28° to 30°C), and thus the sea breeze commences at a time expected in pure sea-breeze events
(Bastin et al. 2005b; Bastin and Drobinski 2005a). The
presence or absence of the mistral blowing in some
regions of the investigated target area contributes to
the time and space inhomogeneity of the sea breeze,
whereas in pure sea-breeze events, the sea-breeze onset
occurs at the nearly same time everywhere along the
coastline [see examples for the 25 and 26 June 2001
sea-breeze events in Bastin et al. (2005b)]. One can also
note the asymmetry of the mistral structure in the
Rhône Valley. The mistral sticks to the eastern flank of
the Rhône Valley whereas it detaches from the Massif
Central after the maximum constriction near Valence
(see Fig. 1b), when it takes a cyclonic curvature due to
the Genoa cyclone (or at least to the leeside trough
downstream the Alps).
At 1100 UTC (Fig. 3c), the data show a surface temperature gradient near the coastline on the sheltered
western region of the target area (longitude ⯝ 4°E) (the
temperature increases from 28° to 32°C). The sea
breeze penetrates inland over a small horizontal range.
The sea-breeze front location is indicated by a thick
dashed line. The location of the sea-breeze front is determined by associating the location of the maximum
temperature and the area where the wind reverses from
the south (sea-breeze flow) to the north (mistral flow).
One can also notice that the northwesterly mistral flow
descends the slope of the Massif Central in its southern
part whereas a wake trailing downstream the Massif
Central, in the north part of the domain, is associated
with very weak winds. The mistral is bounded by the
eastern flank of the Rhône Valley to the east and by the
Massif Central wake to the west at about 4.5°E (Jiang
et al. 2003). In the region where the mistral blows, the
sea breeze cannot penetrate inland.
At 1400 UTC (Fig. 3e), the region where the sea
breeze penetrates now extends farther to the east. In
the western side of the domain, the sea-breeze front has
progressed inland. The maximum penetration of the
sea breeze is observed at about 43.85°N (between 3.8°
and 4.3°E). Between 1400 and 1700 UTC, the seabreeze front progresses inland in the center of the
Rhône Valley (longitude ⯝ 4.5°E). The front line has
“moved” eastward and at 1700 UTC (Fig. 3g), the sea
breeze does not penetrate inland anymore in the western side of the target area, downstream of the Massif
Central. At 1700 UTC, the sea breeze blows in the
Marseille area where it takes a westerly direction because of the coastline shape (Bastin and Drobinski
2005a,b) and because it combines with the mistral. The
observations indicate a large region in the Rhône Valley (between 43.7° and 44.4°N) where there is no temperature gradient and where the sea breeze and the
mistral collide (at 43.8°N).
2) VERTICAL
STRUCTURE
The radiosonde launched from Lyon on 22 June 2001
at 1100 UTC shows the synoptic northwesterly flow
impinging on the Alps (Fig. 4, upper row). Below the
potential temperature inversion at 1.6 km, the mistral
blows at about 4.2 m s⫺1 with a more west-northwest
direction (about 310°). Between 1.6 and 2.4 km ASL,
the wind speed increases: it is about 10 m s⫺1 at 2-km
height and takes a constant value of about 13 m s⫺1
above about 2.5-km height. The radiosounding
launched from Nîmes at 1100 UTC (Fig. 4, lower row)
shows a deeper PBL since the potential temperature
inversion height is about 2.0 km. Above the PBL the
wind speed and direction take constant values equal to
those measured with the radiosounding launched from
Lyon, that is, about 13 m s⫺1 and 315°, respectively.
Below the PBL top, the wind is slightly weaker than the
upstream wind speed in Lyon (2.5 m s⫺1 at Nîmes versus 4.2 m s⫺1 at Lyon) ands veers to the northeast below 500-m height. Looking at Fig. 3c, Nîmes is located
in the Massif Central wake, which explains the weakness of the wind speed.
Figures 5 and 6 display the time series of the vertical
profiles of the horizontal wind speed and direction
measured by the UHF wind profilers (top row) and
simulated by Méso-NH (bottom row) on 22 June 2001
at Saint Chamas (STC) and Marignane (MAR), and
Marseille (MRS) and Aix-en-Provence (AIX), respectively (see locations of the UHF wind profilers in Fig.
1c). The UHF wind profilers give a detailed insight on
the alternation between the sea breeze and the mistral
flow. At Saint Chamas, the onset and breakdown of the
sea breeze are illustrated by the UHF wind profiler
data (Figs. 5a and 5b). The UHF wind measurements
show that the mistral blows from the northwest until
about 1530 UTC. Its intensity decreases at sunrise from
about 10 m s⫺1 down to 5 m s⫺1 due to the convection
1654
MONTHLY WEATHER REVIEW
VOLUME 134
FIG. 4. Vertical profiles of (a), (d) wind direction, (b), (e) wind speed, and (c), (f) potential temperature from the 1100 UTC radiosoundings launched from (top) Lyon and (bottom) Nîmes (solid line) on
22 Jun 2001. The 1100 and 1700 UTC simulated profiles at Lyon are displayed with dashed and dash–dot
lines, respectively, in the top row. The 1100 UTC simulated profile at Nîmes is displayed with dashed line
in the lower row.
that increases friction near the surface. At 1530 UTC,
the low-level wind veers from the northwest to the
west-southwest, indicating the sea breeze onset over
Saint Chamas. The vertical extent of the sea breeze
evolves with time: the sea breeze is about 900 m deep at
1600 UTC and 700 m deep at 1800 UTC. The sea breeze
does not blow after 2200 UTC since after 2200 UTC,
the wind turns to the north. At Marignane located
southeast of Saint Chamas, the UHF data are unreliable near the surface and above 1.7-km height (Figs. 5e
and 5f). However, the reliable measurements show the
same time evolution as in Saint Chamas and take very
similar values, even though the estimation of the accurate onset time for the sea breeze is more difficult in the
absence of near-surface measurements. At Marseille
(Figs. 6a and 6b) and Aix-en-Provence (Figs. 6e and 6f),
the time evolution of the UHF wind profiler data differs from that observed in Saint Chamas and Marignane (which are rural/peri-urban areas). At Marseille,
the wind veers from the northwest to the west at about
1530 UTC. Over Marseille the sea breeze usually takes
a southwesterly direction below 500 m (Bastin and
Drobinski 2005a,b; Lemonsu et al. 2005) and the mistral a northwesterly direction (Drobinski et al. 2005), so
in contrast to what occurs at Saint Chamas and Marignane where the sea breeze and the mistral have opposite direction and collide at a well-marked front, the sea
breeze and the mistral combine near Marseille. This
results in a higher wind speed over Marseille than that
measured at Saint Chamas and Marignane, located
closer to the Rhône Valley axis. During the day the
sea-breeze flow hardly reaches Aix-en-Provence, located 30 km north of Marseille. The UHF wind direction does not show evidence of significant direction
shift despite a weakening of the mistral flow in the
evening of 22 June 2001. But the UHF of Aix-enProvence is located farther inland than the others UHF,
and, if the sea breeze reaches Aix-en-Provence, it is
likely that the depth of the sea breeze would be inferior
to the depth of the sea breeze in Marseille or Saint
Chamas. So, the absence of near-surface measurements
does not allow one to capture the sea-breeze flow if
only it exists.
The spatial variability of the vertical structure of the
JUNE 2006
BASTIN ET AL.
FIG. 5. Time vs height plot of (left) wind direction and (right) wind speed as (top) measured by the
UHF wind profiler and (bottom) simulated with Méso-NH on 22 Jun 2001 at (a)–(d) Saint Chamas
(STC) and (e)–(h) Marignane (MAR) (see STC and MAR in Fig. 1c). White color corresponds to
missing data. Abbreviations are observation (OBS) and model (MOD). The contours in (c) and (g), and
(d) and (h), correspond to the difference between the simulated and measured wind speed and direction,
respectively. The contour intervals (CIs) are 5 m s⫺1 and 25°, respectively.
1655
1656
MONTHLY WEATHER REVIEW
FIG. 6. Same as Fig. 5 but at (a)–(d) Marseille (MRS) and (e)–(h) Aix-en-Provence (AIX) (see MRS
and AIX in Fig. 1c).
VOLUME 134
JUNE 2006
BASTIN ET AL.
sea breeze and mistral flow is accessible with the wind
measurements by the airborne Doppler lidar WIND.
Figure 7 displays the wind field measured along leg
B–A by the airborne Doppler lidar WIND (Figs. 7a and
7b) and the corresponding simulated field at 1700 UTC
(Figs. 7c and 7d). Since the vertical resolution is 250 m,
the Doppler lidar observations do not allow us to document the near-surface flow, and it is thus difficult to
distinguish the sea-breeze circulation. However, one
can see that the near-surface flow has a more pronounced westerly component between 43.4° and 43.8°N
(orange) than north of 43.8°N where the flow blows
from the north (red and blue). This figure shows that a
horizontal shear of the wind appears at 43.8°N up to
1.5-km height, with the wind speed decreasing suddenly
from 11–12 m s⫺1 north of this latitude to about 6–8
m s⫺1. Latitude 43.8°N is the location of the sea-breeze
front detected with the surface stations at 1700 UTC in
Fig. 3. So, even though the wind direction is northwesterly, the sea-breeze flow also affects the mistral in altitude.
4. Model validation
A thorough validation of the model is made in order
to rely on the model outputs to analyze the processes
governing the unsteady and inhomogeneous penetration of the sea breeze in presence of the mistral. In
general, Méso-NH reproduces accurately the flow
structure except for the western Alps wake located too
far to the east (longitude greater than 6°E). The surface
wind and temperature fields are evaluated by comparing the measured 10-m horizontal wind components
and 2-m temperature to their simulated counterparts
interpolated at the location of the meteorological surface stations (Fig. 3, right column). The surface wind
and temperature fields are accurately simulated with
Méso-NH, despite the slight bias in temperature in the
northern part of the Rhône Valley (up to 2 K, see Fig.
3). Quantitatively, Fig. 8 shows the histograms of the
difference between the hourly simulated and measured
10-m zonal and meridional wind components and 2-m
temperature during the 24-h period of 22 June 2001.
The measured 10-m wind speed and 2-m temperature
accuracies are 1 m s⫺1 and about 0.1 K, respectively.
The bias between Méso-NH and the measured wind
speed is ⫺0.14 m s⫺1 for the two horizontal components, which is thus nonsignificant. The standard deviation is also small considering the accuracy of the wind
observations. As for the temperature, the largest discrepancies are found in the steep orography regions
where the height of the topography is not accurately
represented in the model and at night when numerical
1657
diffusion slightly deteriorates the simulation of the
near-surface temperature and the katabatic flows. The
average temperature bias is in part due to the underprediction of the 2-m temperature near Valence at the
maximum constriction of the Rhône Valley.
For the evaluation of the simulated vertical structure
of the sea breeze and the mistral, comparisons are made
with the radiosoundings (Fig. 4), the UHF wind profilers (Figs. 5 and 6), and the airborne Doppler lidar
WIND (Fig. 7). The simulated 1100 UTC sounding over
Lyon is extracted from Méso-NH domain 1 and is plotted with dashed line in Fig. 4. Méso-NH is on average
performing well (at most 2 m s⫺1 difference, less than
20° and 1-K differences on average for the wind direction and potential temperature, respectively). It underestimates the near-surface superadiabatic potential
temperature profile due to the coarser resolution of
domain 1 and to the underestimation of the surface
temperature in the north of the domain (see Fig. 3). It
slightly underestimates the potential temperature inversion (1.5 km versus observed 1.6 km), and the flow
below the inversion is more aligned with the Rhône
Valley axis (north–south orientation). The simulated
vertical profile at 1700 UTC (dash–dot line) shows that
the upstream conditions do not vary much between
1100 and 1700 UTC: the PBL potential temperature
slightly increases with time (it is 297.5 K at 1100 UTC
and 299.5 K at 1700 UTC) but the PBL depth remains
similar; the wind slightly veers from the northnorthwest to the north below the potential temperature
inversion height and the wind speed remains constant.
The vertical profile obtained from the radiosounding
launched from Nîmes at 1100 UTC is also accurately
reproduced by the model (domain 2) with at most 2
m s⫺1 difference, less than 20° and 1-K differences on
average for the wind direction and potential temperature, respectively. At Nîmes also the potential temperature inversion height is slightly underestimated by the
model (1.8 km versus observed 2.0 km). The comparison of the wind speed and direction as a function of
time and height measured by the UHF wind profilers
and simulated with Méso-NH is also accurate (Figs. 5
and 6). The average difference between the measurements (having an accuracy of 1–2 m s⫺1) and the simulation does not exceed 3 m s⫺1 and 20° for wind speed
and direction, respectively. The largest discrepancies
(which can reach 5 m s⫺1 and 50° for wind speed and
direction, respectively, at Marignane and Aix-enProvence, particularly) are often found during fast transitions when a slight time lag in the simulation generates large differences. However the differences are very
limited in time and space and thus do not question the
quality of the simulation. Finally, Fig. 7 shows that
1658
MONTHLY WEATHER REVIEW
FIG. 7. Horizontal wind field along leg B–A shown in Fig. 1b. (a) The wind speed and (b) direction
measurements by the airborne Doppler lidar WIND between 1635 and 1653 UTC, and (c) the wind
speed and (d) direction simulated with Méso-NH at 1700 UTC. White color corresponds to missing data
and the acronyms OBS and MOD stand for observation and model, respectively. The contours in (c) and
(d) correspond to the difference between the simulated and measured wind speed (CI 5 m s⫺1) and
direction (CI 25°), respectively.
VOLUME 134
JUNE 2006
BASTIN ET AL.
1659
FIG. 8. Histograms of the difference between the hourly simulated and measured 10-m (a) zonal and
(b) meridional wind components and (c) 2-m temperature during the 24-h period of 22 Jun 2001. The
measured 10-m wind speed and 2-m temperature accuracies are 1 m s⫺1 and about 0.1 K, respectively.
Méso-NH reproduces well the vertical structure of the
flow along the section B–A flown by the airborne
Doppler lidar. The location of the simulated sea-breeze
front is very slightly shifted to the south by about 10 km
with respect to the observations (43.7°N in the simulation versus 43.8°N in the observations) (the largest discrepancies are thus found in this area). The observed
mistral flow is also slightly stronger below 1.5 km AGL
than in the simulation (the difference is about 2 m s⫺1
for a measurement accuracy of about 1 m s⫺1).
In the following, we consider that the Méso-NH
simulation is validated and is used to provide the threedimensional environment necessary for the interpretation of the measurements.
5. Origin of the unsteadiness of mistral–sea-breeze
interaction
Despite very similar conditions upstream of the Alps
(see the similar vertical profiles at Lyon at 1100 and
1700 UTC in Fig. 4) and upstream of the Massif Central
(not shown), Fig. 9, which displays a horizontal cross
section of the simulated wind field at 400 m AGL in the
larger domain (domain 1 in Fig. 1b), shows that the flow
structure in the Rhône Valley delta, south of the maximum constriction near Valence, differs significantly between 1100 and 1700 UTC. Indeed, at 1100 UTC, the
northwesterly synoptic wind impinges on the Alpine
ridge near Lyon and is deflected in the Rhône Valley
where it veers to the north because of channeling and is
accelerated due to valley constriction (Drobinski et al.
2001a). After the maximum constriction near Valence,
the Rhône Valley can be split into two regions: (i) the
western side of the Rhône Valley near the Massif Central, where the wind is weak and is associated with the
wake trailing downstream the Massif Central and which
defines the western boundary of the mistral (Jiang et al.
2003). South of the Massif Central wake, the Tramontane (considered as the companion of the mistral since
they have the same synoptic origin and often blow simultaneously) blows through the Garonne gap between
the Massif Central and the Pyrénées (Fig. 1a) (Georgelin and Richard 1996; Drobinski et al. 2001b); (ii) the
eastern side of the Rhône Valley near the western Alps,
in which the mistral blows and sticks to the valley flank
until the flow detaches from the topography at 6°E lon-
1660
MONTHLY WEATHER REVIEW
VOLUME 134
FIG. 9. Simulated 400-m AGL horizontal wind over the Rhône Valley delta at (a) 1100 and (b) 1700
UTC 22 Jun 2001 from model domain 1. Contour interval is 2 m s⫺1 from 0 to 14 m s⫺1. Lines D1, D2,
D3, and D4 in (a) show the location of the vertical sections studied in section 5.
gitude. Flow separation from the sidewall of the western Alps occurs farther to the east in comparison with
the 28 June 2001 mistral event (Drobinski et al. 2005).
One can note the acceleration of the mistral as it
reaches the Mediterranean Sea because of roughness
reduction. At 1700 UTC, Fig. 9b shows that the flow
structure in the Rhône Valley delta, south of the maximum constriction near Valence, differs significantly
from the flow structure at 1100 UTC. The sea breeze
blows onshore over a region that does not extend very
far inland. Within the sea-breeze flow, the wind speed is
weaker than in the mistral flow and its direction varies
depending on the local coastline orientation and on the
combination with the mistral. Downstream of the Massif Central, one can note two different regions: (i) south
of the Rhône Valley maximum constriction, the Massif
Central wake with weak wind is similar to that at 1100
UTC but with a smaller spatial extent; (ii) south of the
Massif Central (about 3.2°E and 43.5°N), the northwesterly impinging air flows over the Massif Central
where the maximum crest height is about 1000-m
height. In the eastern side of the Rhône Valley, the
mistral sticks to the flank of the valley. In contrast to
the situation at 1100 UTC, a sharp deceleration of the
wind from 10–11 m s⫺1 down to 6–7 m s⫺1 is visible
between about 44.0° and 44.2°N, as shown in Fig. 7.
a. Western side of the Rhône Valley
Figure 10 shows a vertical cross section of the alongsection wind field (␷h ⫺ w) and isentropes transecting
the southern part of the Massif Central at 0600, 1100,
1400, and 1700 UTC. The arrows represent the wind
vector (␷h, w) where ␷h is the along-section horizontal
velocity component and w the vertical wind component. The line D1, along which the section is made, is
indicated in Fig. 9a. Downstream of the Massif Central,
Fig. 10 shows that at 0600 UTC, the airflow that descends the Massif Central is blocked by a stagnant cold
air pool at the foothill of the massif. The pressure difference between the upwind and lee sides of the Massif
Central, which forces the mistral flow over the ridge, is
about 1.8 hPa. At 0900 UTC, the mean sea level pressure field, displayed in Fig. 11, still shows no pressure
gradient across the shoreline in the Rhône Valley (Fig.
11a). As the sun rises, radiative heating induces convective mixing, which in turn erodes the cold air pool.
Convective activity contributes to decrease the mean
sea level pressure near the shore on the western side of
the Rhône Valley (Fig. 11b). Convection is facilitated
by the fact that this area is located within the Massif
Central wake similarly to the 28 June 2001 mistral case
(Drobinski et al. 2005), so the mistral does not contribute to the stabilization of the near-surface layer. Convection has two antagonist effects: (i) it promotes seabreeze circulation due to cross-shoreline pressure gradient and (ii) contributes to reinforce the pressure
difference across the Massif Central ridge and thus intensify the descending mistral flow. Indeed the pressure
difference between the upwind and lee sides of the
Massif Central is about 3 hPa at 1100 and 1400 UTC,
which corresponds to a pressure difference increase of
about 1.2 hPa between 0600 UTC and midday. Looking
at Figs. 11b and 11c, one can note that the crossshoreline pressure difference is nearly zero at 0900
UTC and is about 0.9 hPa at 1100 and 1400 UTC. The
deepening of the surface heat low in the southwestern
JUNE 2006
BASTIN ET AL.
1661
FIG. 10. Vertical cross section of the ␷h ⫺ w wind components (␷h is the along-section horizontal wind component
and w the vertical wind component multiplied by a factor 10 for legibility) and the isentropes along section D1
transecting the western part of the Massif Central (see Fig. 9a) at (a) 0600, (b) 1100, (c) 1400, and (d) 1700 UTC
22 Jun 2001. The “negative distance” indicates the distance to the coastline over land, whereas the “positive
distance” indicates the distance to the coastline over sea.
area at the foothill of the Massif Central is thus the
main cause of the increase of the pressure difference
between the upstream and downstream sides of the
Massif Central. At 1100 UTC, the still-weak mistral
flow is disrupted by convection as it descends along the
slope (see Fig. 10b between 40 and 60 km onshore) and
forms a vortex similar to a lee rotor. Farther to the
south a secondary updraft is produced at the sea-breeze
front located at about 43.6°N and 4°E (i.e., 10 km onshore in Fig. 10b) where an adverse pressure gradient is
found (see the 1012.9-hPa isoline in Fig. 11b). Figure
12a displays the same horizontal cross section as Fig. 9a
but for the simulated turbulent kinetic energy (TKE)
field. The contribution of convection to TKE is less
than 1 m2 s⫺3 as also shown in Bastin and Drobinski
(2005a) for the 25 June 2001 pure sea-breeze event. The
regions where the TKE exceeds this value are regions
where mechanical TKE also contributes. We clearly see
on the western side of the Rhône Valley an area of
particularly large TKE values, which corresponds to the
area where the mistral and the sea breeze collide near
the shore. At 1400 UTC, the response of the atmospheric flow due to the pressure difference increase between the upstream and downstream sides of the Massif
Central leads to the strengthening of the mistral flow
and allows the descending mistral flow to penetrate
down to the surface (Fig. 10c). A strong updraft is visible at the sea-breeze front since the adverse pressure
gradient generates boundary layer separation. The sea
breeze has not penetrated much farther inland but its
intensity has increased (from 4 to 6 m s⫺1) due to the
descending flow that has reinforced with time (it blows
at about 9 m s⫺1 at 1400 UTC), enhancing frontogenesis (Arritt 1993). A sea breeze “head” forms at the
front with strong upward motions up to 2 km ASL. The
mistral is thus lifted up and goes on blowing above the
1662
MONTHLY WEATHER REVIEW
VOLUME 134
FIG. 11. Simulated mean sea level pressure (hPa) in the Rhône Valley delta at (a) 0900, (b) 1100, (c) 1400, and
(d) 1700 UTC 22 Jun 2001 from model domain 2.
sea-breeze flow. The mistral acts as a rigid lid that prevents the vertical extension of the sea breeze. Indeed,
on 22 June 2001, the sea breeze is only 500–700 m deep,
whereas on 25 and 26 June 2001, the sea breeze extends
up to about 1200 m (Bastin et al. 2005b; Bastin and
Drobinski 2005a). At 1700 UTC, the pressure difference between the upstream and downstream sides of
the Massif Central increases up to about 4 hPa, leading
FIG. 12. Simulated 400 m AGL turbulent kinetic energy over the Rhône Valley delta at (a) 1100 and
(b) 1700 UTC 22 Jun 2001 from model domain 1. Contour interval is 1 m2 s⫺3 from 1 to 3 m2 s⫺3.
JUNE 2006
BASTIN ET AL.
1663
FIG. 13. Same as Fig. 10 but along section D2 transecting the northern Massif Central (see Fig. 9a) at
(a) 1100 and (b) 1700 UTC 22 Jun 2001.
to the reinforcement of the mistral flow, which in turn
moves the sea-breeze front about 8 km offshore (Fig.
10d; Estoque 1962; Arritt 1993). Similarly to the situation at 1100 UTC, the largest TKE values are found at
the sea-breeze front skirting slightly offshore along the
coast (Fig. 12b). However, the background TKE value
has decreased because of decreasing radiative heating.
Downstream of the northern slopes of the Massif
Central, a persistent region of low wind speed is visible
in Fig. 9 and corresponds to the wake trailing downstream of the Massif Central, which delimits the western boundary of the mistral as shown by Jiang et al.
(2003) and Guénard et al. (2005b, hereafter GDC) (autumn cases) and Drobinski et al. (2005) (summer case).
These studies show that this wake is caused by the occurrence of a hydraulic jump on the leeward side of the
Massif Central. Figure 13 is similar to Fig. 10 at 1100
and 1700 UTC along section D2 transecting the northern part of the Massif Central and indicated in Fig. 9a.
It shows evidence of the hydraulic jump occurrence associated with flow deceleration from 7 to 5 m s⫺1. At
1100 UTC, the wake has a large horizontal extension
and is found offshore, at about 43.2°N and 4°E immediately to the north of the Tramontane flow, similar to
the cases studied by Jiang et al. (2003), GDC, and
Drobinski et al. (2005). Figure 12a confirms hydraulic
jump occurrence since it shows that along the northern
slopes of the Massif Central, large values of TKE corresponding to the hydraulic jump that contributes to
the wake formation are found at the same location
where the hydraulic jump is diagnosed by Drobinski et
al. (2005) for the 28 June 2001 mistral case (i.e., along
section 44.5°N, 3.8°E– 45°N, 4.2°E). At 1700 UTC, the
horizontal extension of the wake has considerably decreased as also shown by much weaker TKE values
(Fig. 12b). The wake is disrupted by the mistral flowing
over the southwesterly part of the Massif Central (see
Fig. 10d) and blowing closer to the western flank of the
northern part of the Rhône Valley as shown in Fig. 9b.
The mean sea level pressure field shows a surface low
deepening at 1400 and 1700 UTC in the region of the
wake that tends to suck in the mistral flowing along the
eastern flank of the Rhône Valley.
b. Eastern side of the Rhône Valley
In the middle of the Rhône Valley, the delay of the
sea-breeze onset over land is due to (i) the northerly
mistral flow that maintains the sea-breeze front offshore (Arritt 1993) and (ii) the continental colder air
advection by the mistral that partly compensates radiative heating. Figure 14 shows a vertical cross section of
the along-section wind field (␷h ⫺ w) and isentropes
along the Rhône Valley at 1100 and 1700 UTC. At 1100
UTC, Figs. 3c, 3d, 9a, and 14a show that the mistral
accelerates where the valley is the narrowest near Valence. The studies by Pettré (1982), Drobinski et al.
(2005), and Corsmeier et al. (2005) have shown that
conditions are often propitious to hydraulic jump occurrence downstream in the Rhône Valley. In the
present case, despite upstream conditions favorable to
hydraulic jump occurrence [the 1100 UTC radiosounding launched from Lyon and shown in Fig. 4 gives an
upstream Froude number (ratio between the velocity of
the fluid and the velocity of the gravity wave at the base
of the inversion layer) of 0.28, which is the minimum
value favorable to hydraulic jump occurrence; see
Pettré 1982], Fig. 14a does not show evidence of flow
deceleration associated with PBL deepening which is
the main feature of hydraulic jump occurrence. A reason why the flow does not transition to subcritical regime is that the wake trailing downstream of the Massif
1664
MONTHLY WEATHER REVIEW
VOLUME 134
FIG. 14. Same as Fig. 10 but along the Rhône Valley at (a) 1100 and (b) 1700 UTC 22 Jun 2001.
Central acts as a virtual sidewall that channels the mistral flow, thus preventing flow deceleration (Fig. 9a).
Indeed at 1100 UTC, Fig. 11b shows that, at the exact
location of the Massif Central wake, the mean sea level
pressure is constant and equal to 1012.6 hPa. East of the
eastern wake boundary, the isobars show a continuously increasing along-wind pressure gradient in the direction of the southeast along the eastern flank of the
Rhône Valley, evidencing the acceleration of the flow
between the maximum constriction and the exit of the
Rhône Valley. Figure 12a shows large TKE values in
the middle and eastern side of the Rhône Valley due to
shear production of TKE within the mistral flow (Caccia et al. 2004). On the contrary, at 1400 and 1700 UTC,
Fig. 11d shows that convection contributes to decrease
the mean sea level pressure in the middle of the Rhône
Valley: the mean sea level pressure minimum is located
at 43.8°N at 1400 UTC (1011.7 hPa) and at about 44°N
at 1700 UTC (1011.1 hPa). This mean sea level pressure
minimum contributes to sea-breeze inland penetration
and strong frontogenesis as the sea breeze collides with
the mistral. The stably stratified sea-breeze flow (Fig.
14b) is associated with absence of TKE production, and
a distinct separation is visible between the sea breeze
and the mistral, which is, on the contrary, associated
with large TKE values (Fig. 12b). The adverse pressure
gradient leads to boundary layer separation of the mistral, which is lifted up above the sea-breeze flow. From
a hydraulic point of view, the presence of the seabreeze flow creates the downstream conditions favorable to hydraulic jump occurrence upstream of the seabreeze front at about 44°N since the air is nearly at rest
at the sea-breeze front (Drobinski et al. 2001a). The
strength of the mistral flow prevents the sea-breeze
flow from penetrating very far inland: in the present
case, the maximum inland penetration in the middle of
the Rhône Valley is about 40 km at 1700 UTC and is to
be compared to the 100-km inland penetration on 25
June 2001 (when a weak northerly blows within the
Rhône Valley) and to the 150-km inland penetration on
26 June 2001 (in presence of a slight prevailing southerly flow) (Bastin et al. 2005b).
Farther to the east, near Marseille, the sea breeze
penetrates late in the afternoon at about 1500 UTC. On
25 June 2001, the sea breeze starts blowing over
Marseille at 0800 UTC, and on 26 June 2001 at 1000
UTC (on this day, the prevailing onshore flow tends to
delay the land–sea temperature contrast by advecting
cold marine air over land; Bastin et al. 2005b). Figure 3
shows that at 1700 UTC, the sea breeze penetrates on a
25-km horizontal range. Figure 15, which displays a vertical cross section of the along-section wind field (␷ h ⫺
w) and isentropes transecting the western Alps at 1700
UTC along line D3 (Fig. 9a), shows that the late and
FIG. 15. Same as Fig. 10 but along section D3 transecting the
western Alps over Marseille (see Fig. 9a) at 1700 UTC 22 Jun
2001.
JUNE 2006
BASTIN ET AL.
1665
FIG. 16. Simulated mean sea level pressure (hPa) throughout the domain 1, on 22 Jun 2001 at (a)
1100 and (b) 1700 UTC 22 Jun 2001.
limited penetration range is caused (i) by the mistral,
which is most intense in this area (see also Fig. 9b), and
thus prevents sea-breeze penetration, and (ii) by the
local topography that can reach several hundreds of
meters around Marseille, which is another obstacle to
sea-breeze penetration.
The sea-breeze flow also has an impact on the structure of the mistral flow along the eastern flank of the
Rhône Valley. Indeed, the sea-breeze front acts as a
barrier that redirects the mistral and enhances its cyclonic curvature as it exits the Rhône Valley. The mistral
is thus confined between the sea-breeze front and the
western Alps flank where it reaccelerates. Indeed, the
along-wind mean sea level pressure gradient in the easternmost part of the domain shown in Fig. 11 is equal to
1.5 ⫻ 10⫺2 hPa km⫺1 at 0900 UTC, 2.25 ⫻ 10⫺2 hPa km⫺1
at 1100 UTC, 3.0 ⫻ 10⫺2 hPa km⫺1 at 1400 UTC, and
3.75 ⫻ 10⫺2 hPa km⫺1 at 1700 UTC. It demonstrates
the effect of constriction by the sea breeze, which
“squeezes” the mistral flow between the sea-breeze
front and the eastern flank of the Rhône Valley. The
wake trailing downstream the western Alps east of 6°E
longitude with flow reversal (Fig. 9a) is due to flow
separation caused by dissipation in an oblique hydraulic
jump (or flank shock; Schär and Smith 1993). The oblique jump is easily discernable since at the location
where it occurs, a maximum of turbulent kinetic energy
is created (Fig. 12). However, despite the very persistent pattern of the TKE field in this area, Fig. 16, which
displays the mean sea level pressure field of model domain 1, shows that downstream flow separation the
mean sea level pressure field evolves substantially between 1100 and 1700 UTC. At 1100 UTC, the downstream mean sea level pressure pattern is very similar to
that of a pure mistral event such that of 28 June 2001
(Drobinski et al. 2005). At 1700 UTC, convection over
land and the associated sea-breeze flow in the western
Alps wake region (region of no TKE) contributes to
distort the mean sea level pressure field. This “seabreeze air mass” forms a secondary pressure low visible
at 43.3°N and 7°E in Fig. 16b, which creates a secondary
cyclone that maintains the mistral away from the Alps
flank. In the afternoon, the origin of flow separation
thus differs from that discussed by Drobinski et al.
(2005) for the 28 June 2001 mistral event. During the 28
June 2001 mistral event, the Alpine wake is caused by
a combination of flow separation and hydraulic-jumpinduced dissipation of the flow descending the Alps
slope from the north. In the present case, the Alpine
wake is caused by oblique hydraulic-jump-induced flow
separation reinforced by the presence of sea-breeze airmass downstream flow separation occurrence. In addition, the downslope wind blowing along the southern
slope of the Alps does not experience any hydraulic
jump: in fact, Fig. 17, which displays a vertical cross
section of the along-section wind field (␷ h ⫺ w) and
isentropes transecting the western Alps at 1100 and at
1700 UTC along section D4 (Fig. 9a), shows no evidence of hydraulic jump along the Alps slope and advocates for a disconnection between the low-level wake
(below 1 km ASL) induced by flow separation and the
descending air masses that flow over the recirculating
flow. Compared to the 28 June 2001 mistral case at
approximately the same time, the mistral sticks to the
Alps sidewalls much longer on 22 June 2001. Two reasons can be invoked: (i) the mistral wind is much
weaker; (ii) the upstream stagnation point is located
more to the south than for the 28 June 2001 mistral
1666
MONTHLY WEATHER REVIEW
VOLUME 134
FIG. 17. Same as Fig. 10 along section D4 transecting the western Alps (see Fig. 9a) at (a) 1100 and
(b) 1700 UTC 22 Jun 2001.
case, because of the northwest direction of the impinging flow.
6. Summary
This study has pointed out the complexity of the
structure of the mistral and the sea breeze on 22 June
2001. Indeed, it is shown that on 22 June 2001, the sea
breeze displays a large horizontal heterogeneity that is
not present on pure sea-breeze events (Bastin et al.
2005b). The sea-breeze asymmetry between the western and eastern sides of the Rhône Valley and its unsteadiness are highly correlated with the different
mechanisms that drive the mistral flow over these two
regions. We find the following:
• West of the domain (downstream of the Massif Cen-
tral), the sea breeze commences at about 1000 UTC.
The sea-breeze onset is facilitated by the wake trailing downstream the massif due to dissipation in hydraulic jump occurring on the leeward slope of the
Massif Central (corresponding to large mechanical
production of turbulent kinetic energy). Later in the
afternoon, the deepening of the leeward surface low
due to radiative-heating-induced convection favors
the intensification of the mistral that descends the
Massif Central slope and stops the sea-breeze inland
propagation. At 1700 UTC, the mistral pushes the
sea-breeze front offshore, marking the cessation of
the sea breeze in this area. Strong mixing evidenced
by large turbulent kinetic energy value is found at the
sea-breeze front.
• In the middle of the Rhône Valley, the mistral experiences acceleration at the maximum construction
near Valence as generally seen in previous studies
(Pettré 1982; Drobinski et al. 2005; Corsmeier et al.
2005). At 1100 UTC, the Massif Central wake acts as
a virtual sidewall that forces the channeling of the
mistral, which continuously accelerates along the
Rhône Valley (the along-wind mean sea level pressure gradient increases in the direction of the south).
At 1700 UTC, convection has contributed to form a
pressure minimum in the middle of the Rhône Valley. This has two antagonist effects: (i) it promotes
sea-breeze circulation due to cross-shoreline pressure
gradient and (ii) contributes to intensify the mistral,
which continues to accelerate south of the maximum
constriction until it collides with the sea-breeze front
over land about 40 km from the shore (i.e., about
44°N) inducing a hydraulic-jump-like boundary layer
separation. Because of the mistral blowing from the
north and advecting a continental colder and dryer
air mass (Bastin et al. 2005a), the sea breeze takes
longer to penetrate inland and the inland propagation hardly reaches 40 km from the shore.
• East of the domain, the sea-breeze front acts as a
barrier that reinforces and redirects the mistral flow,
enhancing the cyclonic curvature of the mistral when
it exits the Rhône Valley. The mistral is thus confined
between the sea-breeze front and the western Alps
flank, as revealed by the mean sea level pressure pattern. At about 6°E longitude, flow separation, associated with energy dissipation in an oblique jump,
occurs inducing a wake associated with flow reversal
similar to the 28 June 2001 mistral case (Drobinski et
al. 2005), in good agreement with the prediction of
Schär and Smith (1993) theory. The oblique jump is
easily discernable since at the location where it occurs, a maximum of turbulent kinetic energy is created. Despite the very persistent structure of the oblique jump, sea-breeze flow development downstream of flow separation also contributes to
JUNE 2006
TABLE 2. Sea-breeze main characteristics in the Marseille area
in the sea-breeze-only situation (from Bastin et al. 2005b) and in
the combined sea breeze and mistral situation.
Feature
Sea-breeze onset
Sea-breeze depth
Sea-breeze direction
Sea-breeze intensity
Inland penetration
1667
BASTIN ET AL.
Sea breeze only
Between 0800 and
1000 UTC
Up to 1500 m
Southwest
About 5 m s⫺1
About 100 km
Sea breeze
and mistral
After 1500 UTC
⬍1000 m
West
About 5 m s⫺1
⬍50 km
maintain the mistral away from the Alps flank in the
afternoon. The air that flows over the western Alps
from the north blows over the western Alps wake,
which prevents mixing with the air coming from aloft.
Finally, the description of the flow blowing in the
region of Marseille allows us to compare the sea-breeze
main features studied by Bastin et al. (2005b) for situations of sea breeze only with those for the present
situation of combined sea breeze and mistral. The summary of this comparison is reported in Table 2. It
clearly shows that the cold and dry mistral flows blowing against the sea breeze (i) delays the sea-breeze onset at Marseille; (ii) limits the vertical and horizontal
extents of the sea breeze; and (iii) redirects the sea
breeze.
Such combined sea-breeze/mistral events generate
the highest pollution levels nearby the emission
sources, which are also the most densely inhabited area
in southeastern France (Marseille has about one million
inhabitants). The highly unsteady and inhomogeneous
structure of the sea breeze at the Rhône Valley exit, as
well as the small inland penetration (less than 40 km)
make difficult the accurate and reliable prediction of
such events. Models with high-resolution mesh grid are
obviously needed to address this issue.
Acknowledgments. This work was conducted at Service d’Aéronomie and Laboratoire de Météorologie
Dynamique of the Institut Pierre Simon Laplace. The
authors thank the two anonymous referees for their
relevant comments, which helped to improve the manuscript significantly; R. Benamara and R. Vautard for
fruitful discussion; M. C. Lanceau for help in collecting
the referenced papers; and B. Cros and P. Durand for
the coordination of the experiment. The flights of the
DLR Falcon were funded partly by the Coordinated
Access to Aircraft for Transnational Environmental
Research (CAATER) program of the European Commission. We would also like to thank E. Nagel (DLR)
for assistance in operation of the WIND system, the
pilots of the DLR flight facility R. Welser and M.
Hinterwaldner, the Falcon technician, and the great
support of the CAATER facilator A. Giez (DLR). In
the framework of the French programs PNCA and
PRIMEQUAL-PREDIT, ESCOMPTE was performed
thanks to funding from the Ministère de l’Écologie et
du Développement Durable (MEDD), the Agence de
l’Environnement et de la Maîtrise de l’Énergie
(ADEME), the Institut National des Sciences de
l’Univers (INSU), Météo-France, the Institut National
de l’Environnement Industriel et des Risques
(INERIS), the German funding agency (BMBF), the
Institute for Meteorology and Climate Research (IMKKarlsruhe), the Joint Research Center (JRC-Ispra), the
Swiss Federal Institute of Technology (EPFLLausanne), the Centre National d’Études Spatiales
(CNES), Électricité de France (EDF), the air quality
agencies Airmaraix and Airfobep, the cities of
Marseilles and Aix-en-Provence, and the county council of the Bouches-du-Rhône. We particularly thank
M. P. Lefebvre who played a crucial role in the delicate
organization of the flights.
REFERENCES
Arritt, R. W., 1993: Effects of the large-scale flow on characteristics features of the sea breeze. J. Appl. Meteor., 32, 116–125.
Bastin, S., and P. Drobinski, 2005a: Sea breeze induced mass
transport over complex terrain in southeastern France: A
case study. Quart. J. Roy. Meteor. Soc., in press.
——, and ——, 2005b: Temperature and wind velocity oscillations
along a gentle slope during sea-breeze events. Bound.-Layer
Meteor., 114, 573–594.
——, C. Champollion, O. Bock, P. Drobinski, and F. Masson,
2005a: On the use of GPS tomography to investigate water
vapor variability during a mistral/sea breeze event in southeastern France. Geophys. Res. Let., 32, L05808, doi:10.1029/
2004GL021907.
——, P. Drobinski, A. M. Dabas, P. Delville, O. Reitebuch, and C.
Werner, 2005b: Impact of the Rhône and Durance Valleys on
sea-breeze circulation in the Marseille area. Atmos. Res., 74,
303–328.
Bechtold, P., E. Bazile, F. Guichard, P. Mascart, and E. Richard,
2001: A mass flux convection scheme for regional and global
models. Quart. J. Roy. Meteor. Soc., 127, 869–886.
Bougeault, P., and P. Lacarrère, 1989: Parameterization of orography-induced turbulence in a meso-beta scale model. Mon.
Wea. Rev., 117, 1872–1890.
Caccia, J.-L., V. Guénard, B. Bénech, B. Campistron, and P.
Drobinski, 2004: Vertical velocity and turbulence aspects
during mistral events as observed by UHF wind profilers.
Ann. Geophys., 22, 3927–3936.
Caya, D., and I. Zawadzki, 1992: VAD analysis of nonlinear wind
fields. J. Atmos. Oceanic Technol., 9, 575–587.
Corsmeier, U., R. Behrendt, P. Drobinski, and C. Kottmeier,
2005: The mistral and its effect on air pollution transport and
vertical mixing. Atmos. Res., 74, 275–302.
Cros, B., and Coauthors, 2004: The ESCOMPTE program. An
overview. Atmos. Res., 69, 241–279.
1668
MONTHLY WEATHER REVIEW
Delbarre, S., and Coauthors, 2005: Ground-based remote sensing
observation of the complex behaviour of the Marseille
boundary layer during ESCOMPTE. Atmos. Res., 74, 403–
433.
Drobinski, P., J. Dusek, and C. Flamant, 2001a: Diagnostics of
hydraulic jump and gap flow in stratified flows over topography. Bound.-Layer Meteor., 98, 475–495.
——, C. Flamant, J. Dusek, P. H. Flamant, and J. Pelon, 2001b:
Observational evidence and modeling of an internal hydraulic jump at the atmospheric boundary-layer top during a Tramontane event. Bound.-Layer Meteor., 98, 497–515.
——, C. Haeberli, E. Richard, M. Lothon, A. M. Dabas, P. H.
Flamant, M. Furger, and R. Steinacker, 2003: Scale interaction processes during the MAP IOP 12 south foehn event in
the Rhine Valley. Quart. J. Roy. Meteor. Soc., 129, 729–753.
——, and Coauthors, 2005: Summer mistral at the exit of the
Rhône Valley. Quart. J. Roy. Meteor. Soc., 131, 353–375.
Estoque, M. A., 1962: The sea breeze as a function of the prevailing synoptic situation. J. Atmos. Sci., 19, 244–250.
Georgelin, M., and E. Richard, 1996: Numerical simulation of
flow diversion around the Pyrenées: A Tramontana case
study. Mon. Wea. Rev., 124, 687–700.
Guénard, V., P. Drobinski, J.-L. Caccia, B. Campistron, and B.
Bénech, 2005a: An observational study of the mesoscale mistral dynamics. Bound.-Layer Meteor., 115, 263–288.
——, ——, ——, G. Tedeschi, and P. Currier, 2005b: Dynamics of
the MAP IOP-15 severe mistral event: Observations and
high-resolution numerical simulations. Quart. J. Roy. Meteor.
Soc., in press.
Jansá, A., 1987: Distribution of the mistral: A satellite observation. Meteor. Atmos. Phys., 36, 201–214.
Jiang, Q., R. B. Smith, and J. D. Doyle, 2003: The nature of the
mistral: Observations and modeling of two MAP events.
Quart. J. Roy. Meteor. Soc., 129, 857–876.
Lafore, J. P., and Coauthors, 1998: The Meso-Nh atmospheric
VOLUME 134
simulation system. Part I: Adiabatic formulation and control
simulation. Ann. Geophys., 16, 90–109.
Lemonsu, A., S. Bastin, V. Masson, and P. Drobinski, 2005: Vertical structure of the urban boundary layer over Marseille
under sea breeze condition. Bound.-Layer Meteor., in press.
Menut, L., I. Coll, and S. Cautenet, 2005: Impact of meteorological data resolution on the forecasted ozone concentrations
during ESCOMPTE IOP2a and IOP2b. Atmos. Res., 74, 139–
159.
Morcrette, J.-J., 1991: Radiation and cloud radiative properties in
the European Centre for Medium Range weather forecasts
forecasting system. J. Geophys. Res., 96, 9121–9132.
Noilhan, J., and S. Planton, 1989: A simple parameterization of
land surface processes for meteorological models. Mon. Wea.
Rev., 117, 536–549.
Pettré, P., 1982: On the problem of violent valley wind. J. Atmos.
Sci., 39, 542–554.
Puygrenier, V., F. Lohou, B. Campistron, F. Saïd, G. Pigeon, B.
Bénech, and D. Serca, 2005: Investigation on the fine structure of sea-breeze during ESCOMPTE experiment. Atmos.
Res., 74, 329–353.
Reitebuch, O., C. Werner, I. Leike, P. Delville, P. H. Flamant, A.
Cress, and D. Engelbart, 2001: Experimental validation of
wind profiling performed by the airborne 10-␮m heterodyne
Doppler lidar WIND. J. Atmos. Oceanic Technol., 18, 1331–
1344.
——, H. Volkert, C. Werner, A. M. Dabas, P. Delville, P. Drobinski, P. H. Flamant, and E. Richard, 2003: Determination
of airflow across the Alpine Ridge by a combination of
Doppler lidar, routine radiosounding and numerical simulation. Quart. J. Roy. Meteor. Soc., 129, 715–727.
Schär, C., and R. B. Smith, 1993: Shallow-water flow past isolated
topography. Part I: Vorticity production and wake formation.
J. Atmos. Sci., 50, 1373–1400.
Werner, C., and Coauthors, 2001: Wind infrared Doppler lidar
instrument. Opt. Eng., 40, 115–125.